Behind the scenes | Advancing vehicle safety with virtual simulations
In V4SAFETY we focus on prospective safety performance assessment using virtual simulations. We provide a framework for conducting computer simulations to assess the long-term performance and impact of road safety solutions. Our simulations include various models that cover both pre-crash and in-crash scenarios. These include pre-crash simulation models, which analyze vehicle dynamics, environmental conditions, and infrastructure; in-crash simulation models, which assess safety features like airbags and seatbelts; and human models, which predict injury outcomes in accidents.
Work Package (WP) 5
The primary goal of the team working on Task 5.2 is to document and consolidate pre-crash models for safety performance assessment. In WP5, V4SAFETY partners have developed two open-source generic models: an AEB Technology model and a Forward Collision Driver model. These models are available to help researchers, students, and newcomers explore prospective safety performance assessments.
AEB Technology Model: enhancing vehicle safety
The AEB (Automatic Emergency Braking) model is designed to improve vehicle safety through a series of interconnected modules:
· Object Sensor Module: this module employs a 2D sensor to detect traffic participants, recognizing them when a specified portion of their bounding box enters the sensor’s field of view (FOV),
· Collision Prediction Module: this module predicts potential collisions using models like Constant Velocity (CV) and Constant Turning Rate and Velocity (CTRV) over a 3-second horizon. It calculates the Time-To-Collision (TTC), and if a collision is not predicted, it returns “infinity” (∞),
· Critical TTC Module: the system compares the TTC to a predefined threshold. If the TTC is below or equal to this threshold, the AEB system is triggered,
· Brake Module: the final module calculates the necessary deceleration using a PI controller, ensuring the braking process follows a specific deceleration profile that includes a delay phase, build-up, and maximum deceleration.
Together, these modules work seamlessly to detect potential collisions and automatically apply the brakes, preventing accidents.
The AEB Technology Model is crucial because AEB systems have been mandatory in all new vehicles since July 2024. Despite widespread implementation, ongoing research aims to enhance these systems’ safety potential and robustness. By providing this system as an open-source model, we aim to reduce time-consuming reverse engineering and direct focus on newer safety system innovations.
1. Illustrative example 1 of the AEB model in use
Illustrative example 2 of the AEB model in use
FCW model: simulating driver response
The Forward Collision Warning (FCW) model simulates how a human driver responds to a forward collision warning, which is essential for preventing accidents. It functions as a pre-crash model, enabling users to define conditions when the FCW system is activated and how a driver’s brake reaction times are influenced. This model accounts for several factors that could impact the driver’s response time, using a base reaction time and applying correction factors to adjust for different influences not captured in the data.
The model includes:
· Predefined reaction times: users can define and adjust reaction times based on the scenario.
· Correction factors: these factors allow customization for specific conditions not addressed by base data.
· Literature and configuration: we provide curated resources and example configurations for realistic use cases.
A sampling program generates human response data, which users can integrate into simulations alongside systems like AEB, ensuring a comprehensive representation of human-vehicle interactions.
Open Science practices
The V4SAFETY models are publicly available on the OpenVT platform.
The OpenVT platform is based on GitLab, an open source repository management and version control system. By design, the platform serves for collaborative development as well as for dissemination. This enables a close dialogue and short feedback loops between users and developers.
Virtual Testing
The OpenVT Organisation (OVTO) hosts the ‘OpenVT’ as a collaborative platform for researchers, industry, and other stakeholders in road safety and virtual testing. OVTO is a politically independent non-profit association. Its ambition is to host open-source projects related to virtual testing and biomechanics, providing open access tools for virtual road safety testing.
OVTO was established as a non-profit spin-off from the EU-funded project VIRTUAL (Horizon 2020), with the intention of making a sustainable future happen for the Open Science community and tools that originated in VIRTUAL. It has since then expanded into other projects and initiatives. OVTO promotes and supports research on virtual testing and computer modeling in traffic safety, aiming to achieve Vision Zero for all road users.
Learn more at OVTO.org!